期刊文献+
共找到108篇文章
< 1 2 6 >
每页显示 20 50 100
基于TZ Normalization规整的话者确认阈值选取 被引量:3
1
作者 刘明辉 陈继旭 +1 位作者 戴蓓蒨 李辉 《数据采集与处理》 CSCD 北大核心 2005年第3期311-317,共7页
针对说话人确认中,各目标话者模型输出评分分布不一致而导致系统确认阈值设置的困难,本文采取了通过评分规整确定系统最小检测代价函数(DCF)确认阈值的方法。在分析了已有的两种评分规整方法Z norm a l-ization和T norm a lization的基... 针对说话人确认中,各目标话者模型输出评分分布不一致而导致系统确认阈值设置的困难,本文采取了通过评分规整确定系统最小检测代价函数(DCF)确认阈值的方法。在分析了已有的两种评分规整方法Z norm a l-ization和T norm a lization的基础上,提出了一种结合两者优点的组合规整方法——TZ norm a lization,并据此给出了一种阈值动态修正方法,有效地提高了系统的性能和阈值选取的鲁棒性。对历年的N IST(手机电话语音)评测语料库进行了实验,表明了该方法的有效性。 展开更多
关键词 说话人确认 评分规整 TZ normalization 最小DCF阈值
在线阅读 下载PDF
基于改进YOLOv7的交通信号灯检测
2
作者 郑岚月 张玉洁 《系统仿真学报》 北大核心 2025年第4期993-1007,共15页
针对通用物体检测算法在信号灯检测方面存在着识别精度较低的问题,提出一种专门针对交通信号灯检测任务的改进的YOLOv7算法。去掉20×20检测尺度,添加160×160的检测尺度,在使模型轻量化的同时增加浅层特征;结合BiFormer中提出... 针对通用物体检测算法在信号灯检测方面存在着识别精度较低的问题,提出一种专门针对交通信号灯检测任务的改进的YOLOv7算法。去掉20×20检测尺度,添加160×160的检测尺度,在使模型轻量化的同时增加浅层特征;结合BiFormer中提出的BRA(bi-level routing attention)与坐标轴注意力,针对交通信号灯位置特点提出了ABRA(axially-guided BRA);针对IoU指标对物体尺寸敏感的问题,引入NWD(normalized wasserstein distance)度量,改进物体定位损失与置信度损失。实验结果表明:改进YOLOv7算法的mAP值达到了97.7%,比原始YOLOv7提高了11.4%,检测速度提升了90帧/s,计算复杂度降低了4.5%。 展开更多
关键词 交通信号灯检测 注意力机制 BiFormer NWD(normalized wasserstein distance) YOLOv7
在线阅读 下载PDF
Normalized Solutions of Nonlinear Choquard Equations with Nonconstant Potential
3
作者 LI Nan XU Liping 《应用数学》 北大核心 2025年第1期14-29,共16页
In this paper,we mainly focus on a type of nonlinear Choquard equations with nonconstant potential.Under appropriate hypotheses on potential function and nonlinear terms,we prove that the above Choquard equation with ... In this paper,we mainly focus on a type of nonlinear Choquard equations with nonconstant potential.Under appropriate hypotheses on potential function and nonlinear terms,we prove that the above Choquard equation with prescribed 2-norm has some normalized solutions by introducing variational methods. 展开更多
关键词 Nonlinear Choquard equation Potential function Variational method Normalized solution
在线阅读 下载PDF
Normalized Positive Ground State Solutions for Nonhomogeneous Kirchhoff Equations
4
作者 ZHANG Xiaocang XU Liping 《应用数学》 北大核心 2025年第3期711-720,共10页
This article studies a class of nonlinear Kirchhoff equations with exponential critical growth,trapping potential,and perturbation.Under appropriate assumptions about f and h,the article obtained the existence of norm... This article studies a class of nonlinear Kirchhoff equations with exponential critical growth,trapping potential,and perturbation.Under appropriate assumptions about f and h,the article obtained the existence of normalized positive solutions for this equation via the Trudinger-Moser inequality and variational methods.Moreover,these solutions are also ground state solutions.Additionally,the article also characterized the asymptotic behavior of solutions.The results of this article expand the research of relevant literature. 展开更多
关键词 Normalized positive ground state solution Nonhomogeneous Kirchhoff equation Variational method Exponential critical growth Trapping potential
在线阅读 下载PDF
Study on Color Difference of Color Reproduction of 3D Objects
5
作者 GU Chong DENG Yi-qiang 《印刷与数字媒体技术研究》 北大核心 2025年第4期33-38,69,共7页
To investigate the applicability of four commonly used color difference formulas(CIELAB,CIE94,CMC(1:1),and CIEDE2000)in the printing field on 3D objects,as well as the impact of four standard light sources(D65,D50,A,a... To investigate the applicability of four commonly used color difference formulas(CIELAB,CIE94,CMC(1:1),and CIEDE2000)in the printing field on 3D objects,as well as the impact of four standard light sources(D65,D50,A,and TL84)on 3D color difference evaluations,50 glossy spheres with a diameter of 2cm based on the Sailner J4003D color printing device were created.These spheres were centered around the five recommended colors(gray,red,yellow,green,and blue)by CIE.Color difference was calculated according to the four formulas,and 111 pairs of experimental samples meeting the CIELAB gray scale color difference requirements(1.0-14.0)were selected.Ten observers,aged between 22 and 27 with normal color vision,were participated in this study,using the gray scale method from psychophysical experiments to conduct color difference evaluations under the four light sources,with repeated experiments for each observer.The results indicated that the overall effect of the D65 light source on 3D objects color difference was minimal.In contrast,D50 and A light sources had a significant impact within the small color difference range,while the TL84 light source influenced both large and small color difference considerably.Among the four color difference formulas,CIEDE2000 demonstrated the best predictive performance for color difference in 3D objects,followed by CMC(1:1),CIE94,and CIELAB. 展开更多
关键词 Color difference formula 3D objects Light source Gray scale Normalized residual sum of squares
在线阅读 下载PDF
Rate-frequency dependent shear behavior of rough rock joint experiencing normal load oscillations
6
作者 DANG Wen-gang LI Xing-ling +1 位作者 TAO Kang FU Jin-yang 《Journal of Central South University》 2025年第5期1873-1886,共14页
Dynamic disturbances certainly reduce shear strength of rock joints,yet the mechanism needs deeper explanation.We investigate the shear behavior of a rough basalt joint by conducting laboratory shear experiments.Const... Dynamic disturbances certainly reduce shear strength of rock joints,yet the mechanism needs deeper explanation.We investigate the shear behavior of a rough basalt joint by conducting laboratory shear experiments.Constant and superimposed oscillating normal loads are applied at the upper block.Meanwhile,the bottom block moves at a constant shear rate.We investigate the shear behavior by:1)altering the normal load oscillation frequency with a same shear rate,2)altering the shear rate with a same normal load oscillation frequency,and 3)altering the normal load oscillation frequency and shear rate simultaneously with a constant ratio.The results show that the oscillating normal load reduces the coefficient of friction(COF).The reduce degree of COF increases with higher shear rate,decreases when increasing normal load oscillation frequency,and keeps constant if the special ratio,v/f(shear rate divided by normal oscillation frequency),is constant.Moreover,we identify a time lag between peak normal load and peak shear load.And the lagging proportion increases with higher shear rate,and decreases with larger static COF.Our results imply that a lower creep rate with a higher normal load oscillation frequency easily destabilizes the creeping fault zones. 展开更多
关键词 normal load oscillation shear rate oscillation frequency friction reduction time lag
在线阅读 下载PDF
Improvement of large-scale-region landslide susceptibility mapping accuracy by transfer learning
7
作者 ZHANG Wen-gang LIU Song-lin +3 位作者 WANG Lu-qi SUN Wei-xin ZHANG Yan-mei NIE Wen 《Journal of Central South University》 CSCD 2024年第11期3823-3837,共15页
Machine-learning methodologies have increasingly been embraced in landslide susceptibility assessment.However,the considerable time and financial burdens of landslide inventories often result in persistent data scarci... Machine-learning methodologies have increasingly been embraced in landslide susceptibility assessment.However,the considerable time and financial burdens of landslide inventories often result in persistent data scarcity,which frequently impedes the generation of accurate and informative landslide susceptibility maps.Addressing this challenge,this study compiled a nationwide dataset and developed a transfer learning-based model to evaluate landslide susceptibility in the Chongqing region specifically.Notably,the proposed model,calibrated with the warmup-cosine annealing(WCA)learning rate strategy,demonstrated remarkable predictive capabilities,particularly in scenarios marked by data limitations and when training data were normalized using parameters from the source region.This is evidenced by the area under the receiver operating characteristic curve(AUC)values,which exhibited significant improvements of 51.00%,24.40%and 2.15%,respectively,compared to a deep learning model,in contexts where only 1%,5%and 10%of data from the target region were used for retraining.Simultaneously,there were reductions in loss of 16.12%,27.61%and 15.44%,respectively,in these instances. 展开更多
关键词 data-limited cases transfer learning landslide susceptibility machine learning normalization based on the parameters of the source domain
在线阅读 下载PDF
Schur Forms and Normal-Nilpotent Decompositions 被引量:1
8
作者 LI Zhen 《应用数学和力学》 CSCD 北大核心 2024年第9期1200-1211,共12页
Real and complex Schur forms have been receiving increasing attention from the fluid mechanics community recently,especially related to vortices and turbulence.Several decompositions of the velocity gradient tensor,su... Real and complex Schur forms have been receiving increasing attention from the fluid mechanics community recently,especially related to vortices and turbulence.Several decompositions of the velocity gradient tensor,such as the triple decomposition of motion(TDM)and normal-nilpotent decomposition(NND),have been proposed to analyze the local motions of fluid elements.However,due to the existence of different types and non-uniqueness of Schur forms,as well as various possible definitions of NNDs,confusion has spread widely and is harming the research.This work aims to clean up this confusion.To this end,the complex and real Schur forms are derived constructively from the very basics,with special consideration for their non-uniqueness.Conditions of uniqueness are proposed.After a general discussion of normality and nilpotency,a complex NND and several real NNDs as well as normal-nonnormal decompositions are constructed,with a brief comparison of complex and real decompositions.Based on that,several confusing points are clarified,such as the distinction between NND and TDM,and the intrinsic gap between complex and real NNDs.Besides,the author proposes to extend the real block Schur form and its corresponding NNDs for the complex eigenvalue case to the real eigenvalue case.But their justification is left to further investigations. 展开更多
关键词 Schur form normal matrix nilpotent matrix tensor decomposition vortex identification
在线阅读 下载PDF
Shear mechanical properties and frictional sliding responses of rough joint surfaces under dynamic normal displacement conditions 被引量:1
9
作者 ZHU Qiang YIN Qian +9 位作者 TAO Zhi-gang HE Man-chao ZHENG Bo-wen JING Hong-wen REN Shu-lin ZHANG Qiang MENG Bo BAI Dong-feng WU Sai-sai WU Jiang-yu 《Journal of Central South University》 SCIE EI CAS CSCD 2024年第7期2393-2410,共18页
A comprehensive understanding of the dynamic frictional characteristics in rock joints under high normal load and strong confinement is essential for ensuring the safety of deep engineering construction and mitigating... A comprehensive understanding of the dynamic frictional characteristics in rock joints under high normal load and strong confinement is essential for ensuring the safety of deep engineering construction and mitigating geological disasters.This study conducted shear experiments on rough rock joints under displacement-controlled dynamic normal loads,investigating the shear behaviors of joints across varying initial normal loads,normal loading frequencies,and normal loading amplitudes.Experimental results showed that the peak/valley shear force values increased with initial normal loads and normal loading frequencies but showed an initial increase followed by a decrease with normal loading amplitudes.Dynamic normal loading can either increase or decrease shear strength,while this study demonstrates that higher frequencies lead to enhanced friction.Increased initial normal loading and normal loading frequency result in a gradual decrease in joint roughness coefficient(JRC)values of joint surfaces after shearing.Positive correlations existed between frictional energy dissipation and peak shear forces,while post-shear joint surface roughness exhibited a negative correlation with peak shear forces through linear regression analysis.This study contributes to a better understanding of the sliding responses and shear mechanical characteristics of rock joints under dynamic disturbances. 展开更多
关键词 dynamic normal displacement shear strength variations phase lag phenomenon STICK-SLIP
在线阅读 下载PDF
基于Copula函数和风速正交分量的风矢量联合概率密度研究 被引量:1
10
作者 程正兵 冀骁文 《计算力学学报》 CAS CSCD 北大核心 2024年第5期942-947,共6页
准确地表示风矢量联合概率密度对于风能的评估和风机结构设计具有重要意义,本文将围绕提升风矢量联合概率密度表示结果准确性展开研究。首先,将风矢量表示为风速正交分量,采用混合Normal分布分别对边缘概率密度进行表示,并提出了估计混... 准确地表示风矢量联合概率密度对于风能的评估和风机结构设计具有重要意义,本文将围绕提升风矢量联合概率密度表示结果准确性展开研究。首先,将风矢量表示为风速正交分量,采用混合Normal分布分别对边缘概率密度进行表示,并提出了估计混合Normal分布参数的方法。然后,基于Copula函数考虑风速正交分量间的相关性,从而得到二者的联合概率密度,其中Copula参数通过最小二乘法估计,紧接着通过雅可比变换得到风矢量联合概率密度。最后,采用印度风能研究所的测点S3和S7的每小时平均风速和风向数据,与以往基于Copula函数和风速、风向变量得到的风矢量联合概率密度结果进行对比。结果表明,与以往方法相比,所提方法得到的风矢量联合概率密度结果准确性有很大的提升,本文可为风能的开采和利用提供理论依据。 展开更多
关键词 COPULA函数 正交分解 风矢量 联合概率密度 混合Normal
在线阅读 下载PDF
主动脉缩窄后远端扩张可能性的血流动力学分析
11
作者 吴嘉慧 陈端端 鲁亦凡 《医用生物力学》 CAS CSCD 北大核心 2024年第S01期258-258,共1页
目的动脉缩窄后远端扩张(PSD)是指在主动脉缩窄(COA)下游区域发生的局部扩张。已有研究分析主动脉缩窄后远端扩张的形成机制。然而,血流动力学的变化如何作用于血管壁使得PSD发生还需要进一步的探讨。本研究的主要目的是利用一系列不同... 目的动脉缩窄后远端扩张(PSD)是指在主动脉缩窄(COA)下游区域发生的局部扩张。已有研究分析主动脉缩窄后远端扩张的形成机制。然而,血流动力学的变化如何作用于血管壁使得PSD发生还需要进一步的探讨。本研究的主要目的是利用一系列不同缩窄和扩张程度下的COA和PSD模型揭示主动脉缩窄后远端扩张处的血流动力学差异,评估PSD的发生可能性和破裂风险。方法计算机断层扫描重建正常人主动脉(Normal)。为了分析PSD形成后的血流动力学差异,修改缩窄或(和)扩张区域获得COA和PSD模型。进行血流动力学模拟,计算出动压、涡度、横向壁面剪切应力(Trans WSS)等血流动力学参数。通过修改最小缩窄径、缩窄长度、最大扩张径获得不同缩窄和扩张程度的COA和PSD模型,评估不同程度COA和PSD患者远端(再)扩张的风险。结果高动压、高涡度和高Trans WSS出现并聚集在COA缩窄段远端的凸侧,且在PSD中均有所缓解。动压、涡度、横向壁面剪切应力在远端的均值的顺序均为:COA>PSD>Normal。最小缩窄径越小,均值越大。而不同缩窄长度的均值相差不大。结论缩窄段远端的凸侧受异常血流动力学的影响最大,是扩张形成的主要区域,其受到的高动压、高涡度和高Trans WSS是远端(再)扩张的风险因素。最小缩窄径越小,远端(再)扩张的风险越大。 展开更多
关键词 作用于血管 主动脉缩窄 血流动力学参数 扩张程度 远端的 计算机断层扫描 壁面剪切应力 NORMAL
在线阅读 下载PDF
Normal mouse serum alleviates radiation pneumonitis in mice by inhibiting the focal adhesion signaling pathway
12
作者 YUAN Tong GUO Yuying +1 位作者 ZHANG Junling FAN Saijun 《南方医科大学学报》 CAS CSCD 北大核心 2024年第5期801-809,共9页
Objective To evaluate the therapeutic effect of normal mouse serum on radiation pneumonitis in mice and explore the possible mechanism.Methods Mouse models of radiation pneumonitis induced by thoracic radiation exposu... Objective To evaluate the therapeutic effect of normal mouse serum on radiation pneumonitis in mice and explore the possible mechanism.Methods Mouse models of radiation pneumonitis induced by thoracic radiation exposure were given intravenous injections of 100μL normal mouse serum or normal saline immediately after the exposure followed by injections once every other day for a total of 8 injections.On the 15th day after irradiation,histopathological changes of the lungs of the mice were examined using HE staining,the levels of TNF-α,TGF-β,IL-1αand IL-6 in the lung tissue and serum were detected using ELISA,and the percentages of lymphocytes in the lung tissue were analyzed with flow cytometry.High-throughput sequencing of exosome miRNA was carried out to explore the changes in the signaling pathways.The mRNA expression levels of the immune-related genes were detected by qRT-PCR,and the protein expressions of talin-1,tensin2,FAK,vinculin,α-actinin and paxillin in the focal adhesion signaling pathway were detected with Western blotting.Results In the mouse models of radiation pneumonitis,injections of normal mouse serum significantly decreased the lung organ coefficient,lowered the levels of TNF-α,TGF-β,IL-1αand IL-6 in the serum and lung tissues,and ameliorated infiltration of CD45^(+),CD4^(+)and Treg lymphocytes in the lung tissue(all P<0.05).The expression levels of Egfr and Pik3cd genes at both the mRNA and protein levels and the protein expressions of talin-1,tensin2,FAK,vinculin,α‑actinin and paxillin were all significantly down‑regulated in the mouse models after normal mouse serum treatment.Conclusion Normal mouse serum ameliorates radiation pneumonitis in mice by inhibiting the expressions of key proteins in the Focal adhesion signaling pathway. 展开更多
关键词 normal mouse serum radiation pneumonitis EXOSOMES focal adhesion pathway
在线阅读 下载PDF
Nonparametric Estimation of the Trend Function for Stochastic Processes Driven by Fractional Brownian Motion of the Second Kind
13
作者 WANG Yihan ZHANG Xuekang 《应用数学》 北大核心 2024年第4期885-892,共8页
The present paper deals with the problem of nonparametric kernel density estimation of the trend function for stochastic processes driven by fractional Brownian motion of the second kind.The consistency,the rate of co... The present paper deals with the problem of nonparametric kernel density estimation of the trend function for stochastic processes driven by fractional Brownian motion of the second kind.The consistency,the rate of convergence,and the asymptotic normality of the kernel-type estimator are discussed.Besides,we prove that the rate of convergence of the kernel-type estimator depends on the smoothness of the trend of the nonperturbed system. 展开更多
关键词 Nonparametric estimation Fractional Brownian motion Uniform consistency Asymptotic normality
在线阅读 下载PDF
Exact MAX tests in genetic association analysis
14
作者 Yi Li Jianan Tian +2 位作者 Chenliang Xu Yaning Yang Min Yuan 《中国科学技术大学学报》 CSCD 北大核心 2024年第12期33-38,I0005,I0009,共8页
The three common genetic models(or modes of inheritance)in association analysis are the dominant,additive,and recessive models.It is known that the Cochran-Armitage trend test(CATT)which correctly incorporates informa... The three common genetic models(or modes of inheritance)in association analysis are the dominant,additive,and recessive models.It is known that the Cochran-Armitage trend test(CATT)which correctly incorporates information from genetic models,is more powerful than the commonly used Pearson’s chi-square test.However,the true genetic model is usually unknown in practice,and the power of the CAT test could be substantially reduced with a wrongly specified genetic model.To achieve a power that is close to that of a correctly specified CAT test,it is natural to apply trend tests under different possible genetic models and to report the most significant test result.This results in a MAX-type testing procedure,and it was found that this test is usually more powerful than the Pearson’s chi-square test.Although the signi-ficance(i.e.,p value)of the MAX-type test can be accessed by either large sample approximation or permutation methods,requirements for sample size or simulation replicates are demanding with respect to accuracy and efficiency.This paper proposes an approach to calculate the exact p values of MAX-type tests based on the combinatorial counting method.The simulation results show that the exact method is more accurate than the large sample approximation methods and more computationally efficient than the permutation method,and our method can be readily applied to genome-wide association studies(GWASs).The proposed methodis built in an R package,MaXact,which is available at the https://github.com/Myuan 2019/MaXact/. 展开更多
关键词 exact test MAX test normal approximation PERMUTATION
在线阅读 下载PDF
Improved Berry-Esseen Bound for Rademacher Sum
15
作者 MA li YE Liu HAN Xin-Fang 《应用概率统计》 CSCD 北大核心 2024年第6期910-941,共32页
Let X=Σ_(i=1)^(n)a_(i)ξ_(i)be a Rademacher sum with Var(X)=1 and Z be a standard normal random variable.This paper concerns the upper bound of|P(X≤x)−P(Z≤x)|for any x∈R.Using the symmetric properties and R softwa... Let X=Σ_(i=1)^(n)a_(i)ξ_(i)be a Rademacher sum with Var(X)=1 and Z be a standard normal random variable.This paper concerns the upper bound of|P(X≤x)−P(Z≤x)|for any x∈R.Using the symmetric properties and R software,this paper gets the following improved Berry-Esseen type bound under some conditions,|P(X≤x)−P(Z≤x)|≤P(Z∈(0,a1)),∀x∈R,which is one of the modified conjecture proposed by Nathan K.and Ohad K. 展开更多
关键词 sum of i.i.d.random variables standard normal distribution QUANTILE Berry-Esseen upper bound
在线阅读 下载PDF
基于深度信念网络的风电机组齿轮箱故障诊断方法 被引量:17
16
作者 刘秀丽 徐小力 《可再生能源》 CAS 北大核心 2017年第12期1862-1868,共7页
针对风电机组故障诊断中存在的数据量大,提取故障特征困难等问题,结合深度学习理论的强大感知与自我学习能力,提出一种基于深度信念网络的风电机组齿轮箱故障诊断方法。将原始时域信号数据输入深度信念网络进行训练,通过反向微调学习对... 针对风电机组故障诊断中存在的数据量大,提取故障特征困难等问题,结合深度学习理论的强大感知与自我学习能力,提出一种基于深度信念网络的风电机组齿轮箱故障诊断方法。将原始时域信号数据输入深度信念网络进行训练,通过反向微调学习对深度信念网络进行整体微调,提高分类准确性;同时,在训练过程加入Batch Normalization,减少过拟合几率,提高网络的收敛速度。将该方法用于风电机组行星齿轮箱的故障诊断,比DBN和BPNN算法及传统故障诊断方法的准确率更高。 展开更多
关键词 风电机组齿轮箱 深度信念网络 BATCH normalization 故障诊断
在线阅读 下载PDF
基于深度信念网络和多信息融合的复杂机电装备故障诊断方法 被引量:9
17
作者 刘秀丽 徐小力 《机床与液压》 北大核心 2019年第1期151-156,126,共7页
针对复杂机电装备故障诊断中存在的数据量大、提取故障特征困难等问题,结合深度学习理论强大的感知与自我学习能力,提出一种基于深度信念网络和多信息融合的复杂机电装备故障诊断方法。将多个传感器的原始时域信号数据输入深度信念网络... 针对复杂机电装备故障诊断中存在的数据量大、提取故障特征困难等问题,结合深度学习理论强大的感知与自我学习能力,提出一种基于深度信念网络和多信息融合的复杂机电装备故障诊断方法。将多个传感器的原始时域信号数据输入深度信念网络进行训练,通过反向微调学习对深度信念网络进行整体微调,提高分类准确性,同时在训练过程采用ReLu激活函数和加入Batch Normalization,减少过拟合出现概率的同时提高了网络收敛的速度。将此方法运用到复杂数控加工中心刀具的故障诊断中,结果表明该方法相比传统BPNN算法和采用Sigmoid激活函数的深度神经网络算法准确率更高。 展开更多
关键词 深度信念网络 复杂机电装备 ReLu激活函数 BATCH normalization方法 故障诊断
在线阅读 下载PDF
AlexNet改进及优化方法的研究 被引量:31
18
作者 郭敏钢 宫鹤 《计算机工程与应用》 CSCD 北大核心 2020年第20期124-131,共8页
通过对Normalization、优化器、激活函数三方面对AlexNet卷积神经网络进行了改进及优化。针对LRN(Local Response Normalization)不存在可学习参数,提出了用WN(Weight Normalization)来代替LRN,同时将WN置于所有池化层(Pooling layer)之... 通过对Normalization、优化器、激活函数三方面对AlexNet卷积神经网络进行了改进及优化。针对LRN(Local Response Normalization)不存在可学习参数,提出了用WN(Weight Normalization)来代替LRN,同时将WN置于所有池化层(Pooling layer)之后,提高了AlexNet模型训练的准确率;通过对比分析Adam、RMSProp、Momentum三种优化器在不同学习率(Learning rate)下对AlexNet模型训练的影响,并得出了相应的学习率的优化区间,提高了AlexNet在Optimizer的学习率区间选择上的准确性;针对AlexNet中ReLU激活函数存在的部分权重无法更新以及梯度爆炸问题,提出了ReLU6与Swish的融合分段函数算法,提升了AlexNet模型训练收敛速度以及准确率的同时也缓解了过拟合现象的发生。 展开更多
关键词 AlexNet 卷积神经网络(CNN) normalization 优化器 激活函数
在线阅读 下载PDF
基于分水岭变换和图论的图像分割方法 被引量:28
19
作者 冯林 孙焘 +1 位作者 吴振宇 管慧娟 《仪器仪表学报》 EI CAS CSCD 北大核心 2008年第3期649-653,共5页
提出了一种融合分水岭变换和图论的图像分割方法。利用图像的局部灰度信息进行分水岭变换后,将图像分割成多个小区域,再结合各小区域的灰度和空间信息从全局角度用Normalized Cut方法在区域之间进行分割,产生最终的分割结果。实验结果表... 提出了一种融合分水岭变换和图论的图像分割方法。利用图像的局部灰度信息进行分水岭变换后,将图像分割成多个小区域,再结合各小区域的灰度和空间信息从全局角度用Normalized Cut方法在区域之间进行分割,产生最终的分割结果。实验结果表明,该方法可以消除分水岭变换后所产生的过分割现象,是一种有效的图像自动分割方法。 展开更多
关键词 图像分割 分水岭 图论 Normalized CUT
在线阅读 下载PDF
考虑风电功率条件相关性的广义椭球不确定集合建模 被引量:17
20
作者 吴巍 汪可友 +1 位作者 李国杰 葛延峰 《中国电机工程学报》 EI CSCD 北大核心 2017年第9期2500-2506,共7页
近年来,鲁棒优化在电力系统优化调度领域取得了优良的效果。鲁棒优化的一个关键问题是确定风电功率不确定集合,而现有方法难以依据最新风电功率预测信息更新不确定集合。针对这一问题,该文基于条件copula copula模型,构建考虑风电功率... 近年来,鲁棒优化在电力系统优化调度领域取得了优良的效果。鲁棒优化的一个关键问题是确定风电功率不确定集合,而现有方法难以依据最新风电功率预测信息更新不确定集合。针对这一问题,该文基于条件copula copula模型,构建考虑风电功率条件相关性的广义椭球不确定集合。该方法综合考虑了预测误差的时间相关性,非正态的边缘分布特性,以及与风电功率预测值的条件相关性。此外,提出了条件normal copula模型的解析表达式以及相应采样方法。该方法实现了依据最新预测信息更新不确定集合,提高了模型的准确性,降低了集合的保守度。对某风场历史数据进行仿真分析,结果验证了所提算法的有效性,以及预测误差统计特性对不确定集合拟合效果的影响。 展开更多
关键词 风电功率 NORMAL COPULA 条件分布 采样 自适应更新 不确定集合
在线阅读 下载PDF
上一页 1 2 6 下一页 到第
使用帮助 返回顶部